Conferences

An SMT Packaged 4-Channel mm-Wave PA for 5G Applications

5G mobile devices operating at mmWave will incorporate electronic beam steering to address the difficulties of non-line of sight communications at these high frequencies. This requires the use of multiple parallel receive and transmit channels, which must be implemented in a low cost, compact manner. This presentation will describe a 4-channel PA covering the 27.5 to 28.35GHz 5G band designed by PRFI. It was realised as a single chip component on the 0.15µm E-mode GaAs PHEMT process of Sanan IC and is packaged in a standard over-moulded, 5mm x 5mm plastic QFN package. This results in a low cost, SMT compatible component suitable for incorporation into high volume commercial products. Each PA exhibits a gain of 20dB with an RF output power of 24.5dBm at 1dB gain compression with a PAE of 20%. When operated at 7dB back-off to preserve modulation fidelity the third order intercept point is 31dBm

Broadband matching of continuum mode Power Amplifiers using Particle Swarm Optimisation

Even though continuum modes such as class B/J continuum, CCF and CCF-1 offer a large choice of impedances in the design of an amplifier, only those which satisfy the Foster reactance theorem can be utilized in practise. The a priori selection of these impedances has been shown to minimize the search space for the amplifier design, thereby resulting in the simplified amplifier design process [1]. This procedure has been demonstrated for the design of Class B/J continuum and CCF, however, a generic procedure to find the impedances does not exit. In a typical amplifier design methodology, the step to find the matching network of selected impedances, often requires trial and error of several topologies thatmay be avoided by the use of the Real Frequency Techniques. In this work, we combine the steps for selection of impedances and matching network to demonstrate a methodology which finds the optimal impedances in the design space of continuum mode as well as the corresponding ideal matching network of RFPAs simultaneously.
Our methodology relies on the observation that the waveforms in a continuum mode can be represented by a continuum parameter and the S parameters of a matching network by a polynomial; both consisting of an array of numbers (A). For any A, the device can be simulated and the corresponding performance (ex. Efficiency, Output power, ACPR, IMD etc.,) of the amplifier can be obtained. We define this A and the corresponding amplifier performance as an abstract entity called “particle”. To begin with, M particles (?(1),…,?(?),…,?(?)) are initialized randomly. The parameters in the particle form an objective function depending upon the design requirement and the optimization algorithm. Each particle (?(?)) learns or cooperates with other particles to predict the waveforms and the matching network (A) which may improve the objective function and eventually reach the optimal parameter set after several iterations. Evolutionary and/or machine learning algorithms are suited for finding such optimal particle.
The proposed methodology in conjunction with particle swarm optimization (PSO) is applied to the design of a broadband power amplifier in the class BJF-1 continuum [2] to target constant efficiency and output power between 2.5 - 4.6 GHz. For demonstration, we have chosen a 10 W GaN HEMT from Cree, CGH40010F. The device is biased in deep class AB mode (Vdsq=28 V, Idsq=150 mA) and the optimal loadline resistance (Ropt) is found to be 38.1 Ω. The algorithm can predict the ideal topologies of input and output matching. However, these topologies require further optimization while creating the corresponding layout. The resulting ideal networks are further optimized in EM simulations and their measured values show good match with EM simulations. Even though PSO algorithm is used in this work, any of the evolutionary algorithms or machine learning can be used to implement the proposed methodology.
[1] S. Saxena, K. Rawat, and P. Roblin, “Continuous Class-B/J Power Amplifier Using Nonlinear Embedding Technique," IEEE Trans. Circuits Syst. II Express Briefs, vol. 1, no. 1, pp. 1-1, 2016

Cognitive Beamformer Chips with Smart-Antennas for 5G and Beyond: Holistic FDSOI Technology Solutions including ASIC Correlators

This paper presents RFIC MIMO modules integrating beamformer chips (including PAs, LNAs, T/R switches, power combiners/dividers and digital controls) and smart-antenna arrays including down-converters for 5G. Several solutions: 2x2 (16 antennas) and 4x4 (64 antennas) arrays of RFIC beamformer chips are designed, fabricated and experimentally characterized. Wireless 5G links with real-time high data rate video-streaming capabilities are demonstrated using integrated cameras. Field-Field correlation-based over-the-air near-field and far-field test solutions with and without down-conversion are proposed for MIMO/Massive-MIMO phased-array systems both in frequency and time domains. Cognitive BIST solutions are combined with co-array signal-processing. Perspectives for holistic RFSOI technology solutions, including RF-analog-to-digital converters (RF-ADCs) and adaptive body-biasing systems, are drawn in comparison with BiCMOS, GaAs and GaN based technologies.

Development of an Analogue Phased Array Beamformer for Wide Ku/K Band Applications

This paper presents the design and measurement of a wide 16-22 GHz scalable multi-beam phased array analogue beam former. As the current trend of new chipsets which contain ever more RF circuitry continues, the techniques and principles required to use them also become more challenging. This paper looks at the design criteria required to pack huge amounts of RF circuits together and ensure compliance to beam former theory. The use of automated test and measurement is discussed and how the calibration challenges of such a system can be overcome. The paper uses results and experience which came from the development of EECLs scalable multi-beam, 16/64/256 antenna array.

High Efficiency Class E MMIC Oscillator for X-Band Medical Applications- part A

RF and microwave devices have effectively found applications in the field of medicine and electrosurgery due to the non-ionizing nature of RF and microwave energy. Microwave power sources can be combined with suitably designed antenna and applicator structures to produce highly focused and controlled heating on tissue structures; while minimising distributed tissue heating and the effect on neighbouring tissues [1]. This is called Microwave Ablation. Electrosurgical devices operating at C-band and X-band Presented herein is a high efficiency class E oscillator targeted for X-band and Ku-band surgical devices for the treatment of: tumours, gastro oesophageal reflux disorder, and spider veins. The oscillator has been developed using 4x150 μm GaAs pHEMT transistor Presented herein is a high efficiency class E oscillator targeted for X-band and Ku-band surgical devices for the treatment of: tumours, gastro oesophageal reflux disorder, and spider veins. The oscillator has been developed using 4x150 μm GaAs pHEMT transistor from the WIN PP10-15 MMIC process with support from Semtech Limited. Design, EM simulations and layout have been performed in Keysight’s ADS following a three step design procedure 1) Design of class E switching mode amplifier for optimised PAE performance 2) Design of appropriate feedback network according to the terminal voltages and currents 3) Transient analysis and large-signal optimisation of the oscillator The design of class E amplifier involves selection of: optimum load impedance at the fundamental frequency (also the frequency of oscillation) through load pull analysis, and conjugately matched source impedance to drive the transistor in saturation region. Once the amplifier has been optimised for high power and PAE through large signal EM simulations; voltages and currents at the amplifier’s input and output reference planes are obtained. These are then used to design the feedback network required to provide start-up conditions for the oscillations. In the third step, transient analysis is performed and the oscillator is optimised at the design frequency for the required power and efficiency.

Multi-port Amplifier fed Linear Array

The object of this paper is to describe the  possibility of feeding a linear array with a pair of MPAs, so that a variable amplitude taper can be applied to the array.  This may be achieved by switching edge elements off completely or giving them relatively low amplitude, whilst at the same time using all the amplifiers at or near their maximum. Hence the array may be used most efficiently. The aim of this paper is to show that a pair of 8x8 MPAs may feed a linear array. It will also present some predictions of the radiation patterns.

Your recipe of on-wafer calibration for accurate mm-wave characterisation of advanced silicon devices

System calibration is irreplaceable step of the RF device characterisation process. Traditionally, it consists of two tiers: the first is the probe tip calibration performed on a commercially available coplanar calibration substrates. The second-tier is the de-embedding of the device infrastructure parasitic. Is is conducted using a set of wafer-embedded device-specific de-embedding elements, aka de-embedding “dummies”. We will demonstrate that the characterisation of advanced silicon devices requires alternative approach. We will discuss solutions for both metrology-grade and production-suitable calibration methods and will share some experimental results and practical recommendations.